Swswpc-panel-llncs-16-nov-04a.dvi

Do their Approaches Differ in Adding Semantics to Web Jorge Cardoso1, John Miller2, Jianwen Su3, and Jeff Pollock4 1 Department of Mathematics and Engineering Abstract. Since the new terms, "Semantic Web" and "Web services",have been introduced, researchers have followed two different roads. Fol-lowing one road, academia has focused on developing a new set of lan-guages to enable the automation of Web services execution and inte-gration based on the Semantic Web. On the other road, industry hastaken the lead to propose and develop technologies and infrastructuresto support Web services and Web processes without, until recently, pay-ing much attention to semantics. It is fundamental to analyze the trendthat is being followed with regard to the "Semantic Web" and "Webservices". Therefore, two important questions need to be answered: "dothe approaches taken by academia and industry differ in how they addsemantics to Web services?" and "are their efforts converging or diverg-ing?" This paper, based on a panel discussion at an international con-ference on Web services, which consisted of members of both academiaand industry, addresses precisely these two questions.
In July of 2004, a panel was convened to consider a convergence or divergencebetween academic and industrial approaches to adding semantics to Web service and/or Web process descriptions. Everyone agrees that more semantics (or mean-ing) should be added to Web service descriptions. Differences results when thecommunities address the questions of how and how much. How much semantics?Should a Web service operation be given a full semantic specification, say usingoperational semantics [15] or would a functional classification or categorizationsuffice? How machine processable or understandable should the semantics beon the formality vs. informality scale? For example, a complete and formal se-mantic specification is difficult for humans to create or understand. A simpleragreement based approach predicated on standard interfaces (e.g., port types)may be a better short-term solution. It is also possible to move the standardback to an ontological level and then require the parts of a port type to map toan ontology. This approach to interoperability has proved successful in databaseintegration (where each schema is mapped to a common ontology). One couldexpect similar success for Web services, yet the problem is more complicatedsince the description of operations is more complicated than description of dataobjects.
Given the importance and complexity of the issue (adding semantics), it makes sense that the academic and industrial approaches do differ. The industrialapproach should be near-term, practical and with a high probability of success,while academia can afford to be long-term, ambitious and speculative. However,too much divergence may cause a fracture in which industry settles for too littleand academia will design great things that will never be used.
In this paper, we briefly survey the current research and development oc- curring in academia and industry on Semantic Web Services (SWS). The panelconsisted of researchers from both sectors and the paper strives for a balancedtreatment highlighting the strengths of both approaches, analyzing their differ-ences and seeking common ground for future work.
The paper is organized as follows: Section 2 reviews the brief history of attempting to provide semantics for Web services and relates this to the longhistory of attempting this for programs. Issues and directions are discussed aswell as some aspects of current active research projects are highlighted. Section 3parallel section 2, but from an industrial perspective. Because of the complexityof semantics, there are likely to be diminishing returns if too much is added (e.g.,problems with intractability and undecidability as well as too hard to use). Thissection will start with the currently used standards for describing (WSDL 1.1)[22], publishing (UDDI 2.0) [19] and orchestrating (BPEL 1.1) [1] Web servicesand will consider how semantics are and will impact new (e.g., WSDL 2.0 [23]and UDDI 3.0 [20]) as well as future standards. Section 4 attempts to resolvethe differing approaches into a recipe for long-term cooperation and success ofthis most vital new technological area. Finally, section 5 gives a brief summaryof the most important aspects discussed in this paper.
Academic research into Semantic Web Services began with the work of DAML-Sgroup [4]. The idea was to use a formal language to precisely define what a Webservice does. A basic description along these line is provided by the Web ServiceDescription Language (WSDL). WSDL descriptions are rather shallow and focuson operational aspects. As a consequence, these descriptions are inadequate forautomated discovery or composition of Web services. Much richer and deepermachine-processable descriptions are therefore required. The DAML-S (now theOWL-S [13]) group added profile, process and grounding descriptions. A profiledescribes what the Web service does functionally in terms of input (I), output(O), precondition (P), and result (R), the process describes how it is built out ofcomponents and the grounding maps these to WSDL files. Much of the semanticsis captured in the IOPR specifications.
A Web service, as a software component, has one or more operations that can be invoked as well as its own state. An operation may be described byindicating the types of its inputs and outputs, any preconditions required of theinput as well as the results of the operation (either on the state or the outputsproduced). Actually, this goal of specifying what an operation does or, in general,what a process does has a long tradition in Computer Science and includes workin the fields of program methodology, formal programming language semantics,software engineering and software agents. The problems are complex, but thepotential payoff is great.
Besides the major OWL-S project, there are two ongoing projects being de- veloped in the US, the LSDIS METEOR-S project, and in Europe, the DERISWWS project.
The METEOR-S [14] (METEOR for Semantic Web services) project is fo- cused on the usage of semantics for the complete lifecycle of semantic Webprocesses, which represent complex interactions between semantic Web services.
The METEOR-S project targets research on four important areas of the lifecy-cle of semantic Web processes, namely, annotation, discovery, composition, andexecution. For each of the research stages in the lifecycle a framework, infras-tructure or environment has been developed and implemented. The METEOR-S semantic Web Service Annotation Framework (MWSAF) semi-automaticallymarks up Web service descriptions with ontologies. The algorithms developedmatch and annotate WSDL files with relevant ontologies. The METEOR-S WebService Discovery Infrastructure (MWSDI) uses an ontology-based approach toorganize registries, enabling semantic classification of all Web services based ondomains. Each of these registries supports semantic publication of the Web ser-vices, which is used during the discovery process. The METEOR-S Web ServiceComposition Framework (MWSCF) enhances current Web process compositiontechniques by using Semantic Templates to capture the semantic requirementsof the process [3]. The METEOR-S Web Service Dynamic Process Manager(MWSDPM) allows deployment-time and run-time binding of Web services toan abstract process, based on business and process constraints.
DERI [5] is currently working on a project titled Semantic Web enabled Web Services (SWWS). DERI researchers recognize that to use the full potential ofWeb services and the technology around UDDI, WSDL and SOAP, it is indis-pensable to use semantics, since current technologies provide limited support forautomating Web service discovery, composition and execution. Important ob-jectives of the SWWS initiative include providing a richer framework for Webservice description and discovery, as well as, providing scalable Web service medi-ation middleware. Any necessary mediation would be applied based on semanticdata and process ontologies and semantic interoperation.
Aside from investigations on functional descriptions of Web services, there are also work on behavioral descriptions (see [11]). The behavior signature [11]of a service describes how the service can interact with other services. Providingbehavior signatures is critical in service composition. For example, the two inter-acting services may both wait for messages from each other and none of them canthus proceed [6, 7]. It has been argued that Web service composition, automatedor semi-autmated, critically relies on the interaction patterns in the behaviorspecification [9, 10, 21, 2]. A tool WSAT was recently developed for analyzingconversations and Web service bahaviors [7].
Industrial Research and Development on SWS The industrial research related to semantic Web services depends on the ongo-ing development of open standards that ensure interoperability between differentimplementations. Several initiatives have been conducted with the intention toprovide platforms and languages that will allow easy integration of heteroge-neous systems. The standardization efforts for the technologies that underlieWeb services include Simple Object Access Protocol (SOAP) [17], Web ServicesDescription Language (WSDL), Universal Description, Discovery and Integra-tion (UDDI), and process description languages. Several process description lan-guages have been proposed and studied by the industry.
These languages include W3C WS Choreography Group, Business Process Execution Language for Web Services (BPEL4WS, or simply BPEL) (from Mi-crosoft, IBM, BEA), WSCL (from HP), BPML (from Microsoft), WSCI (fromSUN, BEA, Yahoo, and other), XLANG (from Microsoft), and WSFL (fromIBM).
The WSDL is already well established as an essential building block in the evolving stack of Web service technologies, and is being developed and stan-dardized in the W3C’s Web Services Description Working Group. WSDL is aspecification to describe networked XML-based services. It provides a simpleway for service providers to describe the basic format of requests to their sys-tems regardless of the underlying protocol. WSDL is a key part of the effortof the UDDI initiative to provide directories and descriptions of such on-lineservices for electronic commerce and electronic business. WSDL does not, how-ever, support the specification of processes composed of basic Web services norit envision the use of semantics.
In this area, the BPEL4WS, currently has the most prominent status and en- ables defining business processes as coordinated sets of Web service interactions.
The W3C’s Web Services Choreography Working Group also has been charteredto explore this technical area.
All in all, there are few commercial products available that have successfully implemented a semantic layer alongside robust a Web services infrastructure,this despite significant industrial support which exists for standards such asWSDL, BPEL, and UDDI. As has been mentioned, there are two primary con-siderations for semantics with Web services - the process layer and the datalayer. Most enterprise vendors have indeed recognized the importance and valueof semantic metadata for each area, but tend to implement solutions in propri-etary and brittle ways; using their own metadata formats for internal semanticreconciliation.
With regard to the process and orchestration semantics, many vendors seem to be taking a "wait-and-see" approach while the emerging standards converge.
OWL-S, SWWS/WSML, and BPEL each have important strengths to add toan overarching semantic Web services capability. Leadership from DERI and theW3C have each expressed a strong interest in converging the best of each spec-ification - vendors will no doubt wait for this alignment prior to implementingeither on their own.
The hesitation shared by most commercial vendors will not be shared by many industrial research groups - IBM, HP, France Telecom, and Fujitsu haveall applied semantics to Web services for innovative, discovery-driven use cases.
In contrast to "negotiation-style" semantic Web services, there are others who take a "query-driven" approach. In fact, some commercial vendors have begunimplementing semantic layers on top of Web services as a way to issue queries tothem instead of writing more brittle contracts. Annotating Web services usingthe W3C Web Ontology Language (OWL) can make it simpler evolve servicesin dynamic businesses. To do this, modeling tools map ontologies to Web ser-vice WSDL interfaces and a runtime inference engine issues query plans to theunderlying services. This style of semantic query is clearly distinct from process-centric approaches, but both approaches help automate meaningful access tooverly abundant corporate information.
For Web services to become a platform for semantic service oriented comput-ing, academic and industrial researchers will need to create terminologies, tech-nologies, and products that enable sophisticated solution for the advertisement,discovery, selection, composition, and execution of Web services.
Recently, the Semantic Web Services Initiative (SWSI), an initiative of aca- demic and industrial researchers has been composed to create infrastructurethat combines Semantic Web and Web services to enable the automation in allaspects of Web services. In addition to providing further evolution of OWL-S,SWSI will also be a forum for working towards convergence of OWL-S with the products of the SWWS/WSMO [25]/WSML [24]/WSMX [26] research effort,which supplies Web service providers with a core set of constructs for describingthe properties of their Web services in computer-interpretable form. OWL-S willfacilitate the automation of Web service tasks, including automated Web servicediscovery, composition, and execution. The current version of OWL-S builds onthe Ontology Web Language (OWL) recommendation produced by the Web-Ontology Working Group at the World Wide Web Consortium. OWL-S is thefirst well-researched Web Services Ontology, and has numerous users from theacademia.
WSMO is a complete ontology for the definition of Semantic Web Services.
It follows the WSMF as a vision of Semantic Web Services. WSMO itself isdefined using an ontology language based on F-Logic [12]. It contains all conceptsrequired for Semantic Web Services: Ontology, Mediator, Goal, Web ServiceInterface. The WSML is a family of languages that allow Semantic Web servicedesigners to define Semantic Web services in a formal language. The WSMXprovides a standard architecture for the execution of Semantic Web services.
Its architecture is component-based and one possible implementation of Service-oriented Architectures. WSMX itself has execution semantics.
The largest patch of common research ground that industry and academia have to share is simple, or rather, making semantic Web services simpler. As withall semantic technologies, the rigor of expressing semantic Web services metadata(OWL, OWL-S, F-Logic, XML, etc.) with required precision is daunting withoutgood tools. One day analysts will be dragging-and-dropping process diagramsand point-and-clicking ontology mappings. Until then, researchers in industryand academia would be well served to examine modeling heuristics to lowerbarriers for widespread adoption.
The more likely path of common ground will likely be to reach agreement on ontologies for service descriptions, processes, and security. At an even morefundamental level, researchers will have to measure the strengths and limitationsof different representations such as description logics, horn-logic, and F-Logic forthe erent layers of the semantic Web services architecture. In significant ways,the infusion of semantics will alter today’s conceptions of the service-orientedarchitecture paradigm.
Many believe that a new Web will emerge in the next few years, based on thelarge-scale research and development ongoing on the Semantic Web and Webservices. The intersection of these two, Semantic Web services, may prove to beeven more significant. Academia has mainly approached this area from the Se-mantic Web side, while industry is beginning to consider its importance from theWeb services side. Academia started developing semantic-based Web services lan-guages, such as DAML-S (now OWL-S), to enrich the description of Web servicesto facilitate greater automation. The idea was to make explicit the representa-tion of the semantics underlying data, services, and other resources, providing a qualitatively new level of service. Industry was interested in developing an in-frastructure that could allow software applications to be accessed and executedvia the Web based on the idea of Web services. Their efforts resulted in impor-tant, practical, and functional standards such as UDDI, WSDL, SOAP, XLANG,WSFL, WSCI, BPML, BPEL4WS, etc. While the two approaches can be seen asbeing parallel, recently their is some area of convergence. Both academia and in-dustry have realized that for the sake of automation and dynamism in all aspectsof Web services provision, it was indispensable to create an infrastructure thatcombines, at least to some extent, Semantic Web and Web services technologies(Fig. 1). This paper has highlighted some of the contributions of both industryand academia and discussed recent cooperative efforts such as SWSI. Seman-tic Web Service technology’s potential impact makes it essential for further andexpanding cooperative efforts to be pursued in the future.
1. , Business Process Execution Language for Web Services Version 1.1, 05 May 2003, http://www-128.ibm.com/developerworks/library/ws-bpel/.
2. T. Bultan, X. Fu, R. Hull, and J. Su. Conversation specification: A new approach to design and analysis of e-service composition. In Proc. 12th Int. World Wide WebConference (WWW), May 2003.
3. Cardoso, J. and A. Sheth, Semantic e-Workflow Composition. Journal of Intelligent Information Systems (JIIS), Vol. 12, No. 3 (November 2003) pp. 191-225.
4. DAML Services Coalition (alphabetically A. Ankolekar, M. Burstein, J. Hobbs, O.
Lassila, D. Martin, S. McIlraith, S. Narayanan, M. Paolucci, T. Payne, K. Sycara,H. Zeng), "DAML-S: Semantic Markup for Web Services," in Proceedings of theInternational Semantic Web Working Symposium (SWWS), July 30-August 1, 2001.
5. Digital Enterprise Research Institute, http://www.deri.ie/.
6. H. Foster, S. Uchitel, J. Magee, and J. Kramer. Model-based verification of Web service compositions. In Proc. 18th IEEE Int. Conf. on Automated Software Engi-neering Conference, 2003.
7. X. Fu, T. Bultan, and J. Su. WSAT: A tool for formal analysis of Web service compositions. In Proc. of 16th Int. Conf. on Computer Aided Verification (CAV),2004.
8. Gruber, T.R., Toward principles for the design of ontologies used for knowledge sharing. International Journal of Human-Computer Studies, 1995. 43(5-6): pp. 907-928.
9. J. E. Hanson, P. Nandi, and S. Kumaran. Conversation support for business pro- cess integration. In Proc. 6th IEEE Int. Enterprise Distributed Object ComputingConference (EDOC), 2002.
10. J. E. Hanson, P. Nandi, and D. W. Levine. Conversation-enabled Web services for agents and e-business. In Proc. Int. Conf. on Internet Computing (IC-02), CSREAPress, 2002.
11. R. Hull, M. Benedikt, V. Christophides, and J. Su. E-services: A look behind the curtain. In Proc. ACM Symp. on Principles of Database Systems, 2003.
12. M. Kifer, G. Lausen, and James Wu: Logical foundations of object oriented and frame-based languages. Journal of the ACM, 42(4):741-843, 1995.
13. OWL-S: David Martin, Massimo Paolucci, Sheila McIlraith, Mark Burstein, Drew McDermott, Deborah McGuinness, Bijan Parsia, Terry Payne, Marta Sabou,Monika Solanki, Naveen Srinivasan, Katia Sycara, "Bringing Semantics to Web Ser-vices: The OWL-S Approach," Proceedings of the First International Workshop onSemantic Web Services and Web Process Composition (SWSWPC 2004), July 6-9,2004, San Diego, California, USA.
14. METEOR-S: Semantic Web Services and Processes, LSDIS Lab, University of Georgia, http://lsdis.cs.uga.edu/ and http://swp.semanticweb.org/.
15. G. D. Plotkin, "A Structural Approach to Operational Semantics," University of 16. OWL Web Ontology Language Overview, W3C Candidate Recommendation, 18 August 2003, http://www.w3.org/TR/owl-features/ 17. Simple Object Access Protocol (SOAP) 1.1, http://www.w3.org/TR/SOAP/18. Semantic Web Services Initiative (SWSI), http://www.swsi.org/19. UDDI 21. Web Services Conversation Language (WSCL) 1.0, W3C Note, 14 March 2002, 22. Web Services Description Language (WSDL) 1.1, W3C Note, 15 March 2001, 23. Web Services Description Language (WSDL) 2.0, W3C Working Draft, 3 August 2004, http://www.w3.org/TR/2004/WD-wsdl20-20040803/ 24. Web Service Modeling Language (WSML), http://www.wsmo.org/wsml/index.html25. Web Services Modeling Ontology (WSMO), http://www.wsmo.org/26. Web Services Execution Environment (WSMX), http://www.wsmx.org/

Source: http://me.jtpollock.us/pubs/2005.01-swswpc-panel-llncs-16-nov-04a.pdf